Long-term feeding feeding studies with different retinoids to identify the optimum analogue in terms of maximum anticancer activity coupled with minimum toxicity to the animal. Sequential sampling of normal and carcinogen treated animals maintained on retinoid containing diets, to investigate subcellular changes associated with the anticarcinogenic effect. Determination of whether the retinoid prevents the subsequent growth of initiated cells in the tumors by actually reversing the initiating event, or only delays the appearance of tumors by increasing the latent period before neoplastic growth starts. Determination of the minimum length of time over which retinoids must act on an initiated urothelium before their anticarcinogenic effect has become irreversible.